Dear Statalist,
I am estimating a latent class model using Expectation Maximizing Algorithm (Kenneth Train 2008). This requires repeated estimation of a standard Logit (no random parameter) model for each class. A gmnl is used for a similar model in willingness to pay space following the method by (Arne Risa Hole 2010).
I am having a hard time converging. I understand convergence issues have been posted before on Statalist. But my issue is a little different as I want to understand if the problem is caused by large number of people in my data set.
Earlier on, I used an incorrect data set by chance and the model converges with the incorrect data. The incorrect data set had incorrect person id, which had more choices but less people. Once I fixed the wrong data, which now has more people and less choices for each person, the convergence issue starts. I am wondering if the increase in the number of people (from 200 to 1000) is the issue, since the model gets more complex with more people?
Your advice would be much appreciated!
Thank you for your consideration.
Best Regards,
Lulu
I am estimating a latent class model using Expectation Maximizing Algorithm (Kenneth Train 2008). This requires repeated estimation of a standard Logit (no random parameter) model for each class. A gmnl is used for a similar model in willingness to pay space following the method by (Arne Risa Hole 2010).
I am having a hard time converging. I understand convergence issues have been posted before on Statalist. But my issue is a little different as I want to understand if the problem is caused by large number of people in my data set.
Earlier on, I used an incorrect data set by chance and the model converges with the incorrect data. The incorrect data set had incorrect person id, which had more choices but less people. Once I fixed the wrong data, which now has more people and less choices for each person, the convergence issue starts. I am wondering if the increase in the number of people (from 200 to 1000) is the issue, since the model gets more complex with more people?
Your advice would be much appreciated!
Thank you for your consideration.
Best Regards,
Lulu
